Research & Projects

This project explores the relationship between news sentiment and stock market fluctuations, aiming to enhance real-time stock price prediction using machine learning and cloud-based analysis. The motivation behind this work is the increasing influence of news headlines on financial markets, particularly how AI-generated and human-written articles impact investor sentiment. By leveraging natural language processing (NLP) and time-series analysis, this project integrates news sentiment scores with historical stock price data to build predictive models.

The project was developed using Google Cloud and InfluxDB, enabling scalable data storage and real-time processing. This work was completed as part of my independent research in machine learning and finance, with the potential to contribute to better-automated trading strategies and risk management techniques.

Objectives

• Develop a machine learning model incorporating news sentiment analysis to predict stock price fluctuations.

• Integrate historical stock market data with real-time news articles for a comprehensive dataset.

• Implement an efficient cloud-based architecture for data collection, processing, and model training.

• Identify patterns in sentiment trends that correlate with stock performance for major technology companies (Apple, Amazon, Google, Meta, Nvidia).

• Evaluate the model’s accuracy and reliability in forecasting short-term stock movements.

Results

• Built machine learning models that demonstrated an improved ability to predict stock trends when incorporating sentiment analysis compared to price-only models.

• Discovered that certain news sources and specific types of sentiment shifts had a stronger influence on market movement than others.

• Found that while sentiment-based models improve prediction accuracy, market volatility and external factors (e.g., earnings reports, global events) still play a significant role.

This project focuses on automatically identifying the encryption method used in cipher text through supervised machine learning. With the increasing use of cryptographic techniques in cybersecurity, detecting the type of encryption can help in forensic analysis, security assessments, and automated decryption strategies. This project successfully classifies encrypted text with high accuracy by leveraging machine learning algorithms, demonstrating the potential for AI-driven cryptographic analysis.

The project was developed independently, involving the creation of a custom dataset for training and testing. It integrates key machine learning and visualization libraries, such as Pandas, TensorFlow, and scikit-learn, to enhance data processing, model training, and evaluation.

Objectives

• Develop a supervised machine learning model capable of identifying different encryption methods based on cipher text characteristics.

• Create a fully self-made dataset containing diverse encryption techniques for model training and validation.

• Implement and compare multiple machine learning approaches to optimize classification accuracy.

• Utilize data visualization techniques to analyze patterns in encrypted text features.

• Achieve high accuracy in classifying encryption methods, demonstrating machine learning's potential in cryptanalysis.

Results

• Successfully trained and implemented a machine learning model that classifies encryption methods with 99% accuracy.

• Designed and developed a custom dataset to train and evaluate model performance without reliance on external data sources.

• Integrated machine learning and visualization tools to analyze patterns and improve model interpretability.

• Demonstrated that machine learning can effectively distinguish between different encryption methods, offering potential applications in security research, digital forensics, and automated cryptanalysis.

Coming Soon